bloor anti latour

7/27/2019 Bloor Anti Latour

1/32

Pergamon Stud. Hist. Phil. Sci., Vol. 30, No. 1, pp. 81112, 1999 1999 Elsevier Science Ltd. All rights reserved

Printed in Great Britain0039-3681/99 $19.000.00

Anti-Latour

David Bloor*

1. Introduction

Bruno Latour is a vehement critic of the sociology of knowledge in general, and the

Strong Program in particular.1 For those who are familiar with his books Science in

Action (Latour, 1987), The Pasteurization of France (Latour, 1988) and We Have

Never Been Modern (Latour, 1993) the pivotal role played by these criticisms in

Latours writing will be evident. To those who only know his work by repute, or

who have only read the first edition of Latour and Woolgars Laboratory Life

(Latour and Woolgar, 1979), presenting him as a critic of the sociology of knowl-

edge may seem surprising. Latours work and the Strong Program in the sociology

of knowledge are frequently classed together under the label of social constructiv-ism, and this creates the impression that the two enterprises must be fundamentally

similar. This is reinforced by the fact that Latour wants to go further than sociol-

ogists of knowledge, whose work is said to represent something of a half-way

house. He thinks sociologists are insufficiently radical in their critique of science

(Latour, 1992, p. 273). Nevertheless, in reality, the two approaches are deeply

opposed. In Latours eyes the sociology of knowledge has been a failure, and it

* Science Studies Unit, University of Edinburgh, 21 Buccleuch Place, Edinburgh EH8 9LN, UK.Received 24 September 1997; in revised form 28 November 1997.1The traditional stance towards the sociology of knowledge can be called the weak programme.

This involves the idea that socio-psychological causes need only be sought for error, irrationality anddeviation from the proper norms and methodological precepts of science. Apart from this sociologistscan, at best, illuminate the general conditions which encourage or inhibit science. Examples of thisstance are to be found in the work of Lakatos (1971), Laudan (1977) and more recently Haack (1996).Followers of the strong programme, by contrast, argue for the need to explain, in causal terms, allsystems of belief regardless of how the analyst may evaluate them. It should perhaps be stated at theoutset that the causes in question have never been confined to social causes. Such a limitation wouldbe incoherent. Sensory stimulation by objects in the environment always plays a central role. An accountof this approach can be found in Bloor (1976) and Barnes et al. (1996). The all important, but frequentlymisrepresented, symmetry requirement will be discussed in the course of the present paper. For an

account of the epistemological background and consequences of the programme, which includes a dis-cussion of Haacks paper and other recent attacks, see Bloor, forthcoming.

PII: S0039-3681(98)00038-7

81


2/32

82 Studies in History and Philosophy of Science

will continue to fail unless it adopts an entirely new approach which will qualitat-

ively change its character.

In making these claims Latour knowingly aligns himself with a stance in the

sociology of science associated with the work of Robert Merton (see Latour, 1988,

p. 257). Like Karl Mannheim before him, and many others since, Merton felt thatsociological enquiry into the nature of knowledge was bound to be of a limited

character. It was confined to offering a description of the conditions encouraging

or inhibiting the growth of science. At most it could isolate the causes influencing

the direction of enquiry. The process of cognition itself, however, is governed by

methods and criteria which do not derive from, or vary with, our institutions and

conventions. Scientific knowledge answers to nature and reason, not society. Latour

shares this pessimism about the prospects of the sociology of knowledge, but

reaches the conclusion by a different route. He does not want to go back to asserting

the autonomy of reason and nature. Far from it: he believes sociologists are stilltoo much in thrall to such ideas. They must shake off their remaining influence if

they are to make any progress. We must, as he puts it, take one more turn after

the social turn (Latour, 1992, p. 272).

I think Latour is wrong: there is no further turn to be taken. I dont mean that

sociologists of knowledge have completed their task. Of course they have not.

Fundamental ideas still stand in need of refinement, and there is much work to be

done both empirically and theoretically. My point is that Latours ideas do not

represent the way forward. If anything they are a step backwards. To make good

these claims I shall offer a defence of the Strong Program, first formulating Latours

objections and then describing his own proposals for analysing knowledge. I shall

conclude that his criticisms are based on a systematic misrepresentation of the

position he rejects, and that his own approach, in so far as it is different, is unwork-

able.

In order that my aims should not seem too negative, let me indicate the points

of wider interest which will arise in dealing with these criticisms. Latours errors

about the sociology of knowledge derive from his stance towards a very basic

principle which may be called the schema of subject and object. This schemaimplies that knowledge is to be understood in terms of an interaction between an

independent reality, the object of knowledge, and a knowing subject, embodying

its own principles of receptivity. (Typically, though not necessarily, this subject

will be said to construct representations of the object.) Remarkably, Latour wants

the sociologist to reject this schema. Of course, he is not the first to see the subject

object schema as a source of problems, but it must be remembered that there are

many different ways of interpreting it, and many different levels on which it can

be applied. It would therefore be surprising if the subjectobject schema had no

sphere of legitimate application. There is no doubt that it has some intriguinglimitationsI shall introduce one such in a momentbut, unlike Latour, I see no

advantage in adopting a wholesale opposition to it. On the contrary, under certain


3/32

83Anti-Latour

interpretations, I think there is good reason to retain it. I shall therefore take the

opportunity raised by this challenge to the subjectobject schema to rehearse some

of the fundamental, methodological questions in the field, and to address some of

the confusions currently in circulation.2

2. Criticisms of the Strong Program

Latour argues that both sociologists of knowledge, and their previous critics,

have all worked within the framework of the subjectobject polarity. Everyone has

been assuming that knowledge is to be analysed into two ingredients: one furnished

by the object, the other by the knowing subject. Theories of knowledge are just

the stories we tell about how these two supposed ingredients are to be identified,

how they interact, and in what proportions. Some will lay great stress on the com-

plexity of the knowing subjects contributions, others will see it as a passive recep-tacle, or like a blank sheet waiting to be written on. Some accounts of knowledge

will treat the subject as an individual mind, others will identify it as a group or a

culture. Obviously, for a committed sociologist, the ultimate knowing subject will

be social in character, in short, society. Whatever these differences in approach,

whether nativist or empiricist, individualist or collectivist, the overall schema has

always been the same, and the polarity of subject and object has been imprinted

on the subsequent account. We can think of the schema, says Latour, as a line as

shown in Fig. 1.

While most theories of knowledge of this kind apportion influence between sub-ject and object, notice the two extreme positions on the scale. These end-points

suggest the possibility of avoiding an eclectic or dualistic picture. We could aim

at a purely objectivistic theory and try to explain everything (including society) in

terms of nature. On the other hand, we could aim at a purely subjectivist theory and

try to explain everything (including nature) in terms of society. Latour identifies the

Strong Program in the sociology of knowledge as an approach of the latter type.

It works within the subjectobject polarity but occupies one of the extreme pos-

itions open to theories with this structure. As Latour puts it, for the Strong Program,

Society was supposed to explain Nature! (Latour, 1992, p. 278).

Object(Nature)

Subject

(Society)

Fig. 1. The subjectobject schema.

2Valuable and detailed accounts of the stance toward the subjectobject schema in the work ofHusserl, Heidegger and Wittgenstein are to be found in Kusch (1989) and, with special reference to theanalysis of self-consciousness, Tugendhat (1986). Kusch brings out the way in which, e.g., Heideggersopposition to the schema was an expression of his rejection of the individualistic and transcendental

tendencies in Husserl. The relevance of this for my argument is that rejecting an individualistic andtranscendental version of the subjectobject schema doesnt imply rejecting it in every sense. Therecould still be important work for it to do in the context of, say, an anti-individualistic and naturalisticanalysis of knowledge.


4/32


To understand the significance of Latours characterisation I need to explain the

central idea of the Strong Program, and then relate it to his subjectobject axis.

The main feature of the Program is the so-called symmetry postulate. Both true

and false, and rational and irrational ideas, in as far as they are collectively held,

should all equally be the object of sociological curiosity, and should all beexplained by reference to the same kinds of cause. In all cases the analyst must

identify the local, contingent, causes of belief. This requirement was formulated

in opposition to an earlier prevailing assumption, still defended in many quarters,

which has it that true (or rational) beliefs are to be explained by reference to

reality, while false (or irrational) beliefs are explained by reference to the distorting

influence of society. To take an example often used by critics of the sociology of

knowledge, Mendels discovery of the laws of inheritance are explained by his

observations of the plants in his experimental garden. By contrast, the ideology of

MarxistLeninism, the workings of Stalins dictatorship, and the political opportun-ism of certain Soviet agronomists suffices to explain the attractions of the anti-

Mendelian claims of Lysenko (see Medvedev, 1969; Joravsky, 1970). The sym-

metry postulate signals the rejection of this approach, but what is the alternative?

Given Latours location of the Strong Program, at the extreme subjectivist end of

the spectrum, the only alternative open to the sociologist seems to be that of even

handedly forbidding both groups access to the real (Latour, 1996, p.79). The claim

would then be that neither Mendelism nor Lysenkoism had anything to do with

nature. Both would equally have to be seen as mere projections of some constel-

lation of interests or institutionalised modes of thinking: the white screen on which

society projects its cinema (Latour, 1993, p. 53). Symmetry and subjectivism thus

seem to go together.

Latour, rightly, rejects the symmetry principle understood in this way. Despite

its name it is, he says, deeply asymmetrical because it puts all the explanatory

weight on society and none on nature. It doesnt give proper weight to non-social

things and processes, or acknowledge their contribution to our social arrangements.

At first this may look as if Latour wants to mix together ingredients from society,

and ingredients from nature, in the standard way, as if he merely wants to lureus away from the extreme ends of the subjectobject spectrum. But this isnt his

point. He explicitly rejects such eclecticism. His idea is that we must not try to

explain nature in terms of society, or society in terms of nature, nor should we

explain knowledge as a mixture: we must explain both society and nature, at once,

in terms of a third thing or process. Society and nature are, as he puts it, co-

produced (Latour, 1992, p. 287).

In concrete terms, Latour says the attempt to account sociologically for the subt-

lety and richness of scientific results is a hopeless task. In The Pasteurization of

France he rejects the possibility of accounting for Pasteurs discoveries aboutmicrobes by reference to social facts about Pasteur. Such facts are too sparse for

the enterprise to be plausible:


5/32

85Anti-Latour

Conservatism, Catholicism, love of law and order, fidelity to the Empress, brashness,

passionthose are approximately all we get of the social factors acting on Pasteur.

But they are not much if we put on the other side all the scientific work to be

explained. (Latour, 1988, pp. 257258)

There will have been many conservative, patriotic Catholics with character traitsnot unlike Pasteurs, but they didnt discover the anthrax bacillus. Social categories,

it seems, are not discriminating enough for such an ambitious explanatory undertak-

ing.

It is worth emphasising the logical connection between Latours two main points,

that is, between the idea that the Strong Program explains nature by reference to

society, and the idea that the resources of sociology are too crude to account for

the likes of Pasteurs work. The connecting link is that the subtleties of Pasteurs

work come from the detailed character of the observations he makes. If the Strong

Program denies any role to inputs of this kind, and treats Pasteur as responsiveonly to the social influences on him then, Latour concludes, it cant do justice to

the detailed scientific findings. The dream of a sociological explanation of the

content of science is an idle oneas its traditional critics have always said.

Latours remedy for these defects is to propose a new symmetry principle. Call-

ing the symmetry principle of the Strong Program the first such principle, his

second or generalised version is the idea mentioned above, that both nature and

society should be seen as co-produced. Because, on Latours reading, the Strong

Program explained nature in terms of society, there was no way in which agency

could be attributed to things. All agency resides with society. The second symmetry

principle restores agency to things. It allows a truly symmetrical stance from which

to understand the way in which both nature and society are constituted. Only in

this way, implies Latour, can we acknowledge that to make a scientific discovery

is, at the same time, to change society. For Latour, this represents an advance on

what has gone before, allowing us to see changes in science as themselves changes

in society.

The new principle of symmetry, in which the analyst is poised, as it were, above

both nature and society, can be represented by another, vertical, axis on the dia-gram. This will be orthogonal to the original, subjectobject axis and, of course,

nature and society will be distributed symmetrically about it. Following Latour,

we have something as shown in Fig. 2.

How does Latour understand the point I have labelled origin? It has to be said

that his account is obscure, but I shall come back to this. For the moment all we

need to know is that the vertical axis is treated as a measure of stability (Latour,

1992, p. 285). The stability in question is that of the distinction between nature

and society. The idea is that low down on the axis, near the origin, there is no

clear sense attached to the difference between things that are really in nature, andthings which are merely matters of collective belief or opinion. Positions higher

up represent situations where agents treat the demarcation between nature and


6/32


Nature Society

Origin

Fig. 2. Latours second symmetry principle.

society as fixed and settled. The relations between nature and society cannot there-

fore be represented by a rigid polarity because that polarity itself is subject to

variation along the vertical, stability axis. Scientific innovation, for example, is to

be represented as a movement back and forth along the stability axis, as the inno-

vator modifies the shared sense of what things in themselves really exist, out

there in nature. In modifying our ideas the innovator is, in Latours terminology,

making and re-making both nature and society. As he put it:

That there is a history of the things in themselves seems absurd only to those who

want to fix us forever into the boring confrontation between a subject (or a society)

and an object (or a nature). Meanwhile, innovators are constantly crossing the bound-aries between nature and society, and turning our careful distinction between what

has been revealed, what has been discovered and what has been fabricated into a

shambles. (Latour, 1988, p. 262)

It is worth reflecting on the implications of this passage. The reference to things

in themselves is, of course, partly an allusion to Kantian ideas about the noumenal

basis of the subjectobject schema. I think we may also assume it refers to more

common-sense ideas about the independence of the objects of nature from our

ideas about them. We take for granted that trees and rocks, as well as electrons

and bacilli, have long been stable items amongst the furniture of the universe.Putting aside evolution, geological change and the nebular hypothesis, there is a

sense in which they have no history: they are just there, providing a stable

backdrop for the more volatile happenings on the human stage, where ideas change

and theories come and go. For Latour, however, there is a history for things in

themselves, precisely because he sees himself as having left behind the usual

assumptions of transcendence and independence. Belief in that transcendence and

independence is just an expression of the subjectobject schema he wants to reject.

We have now seen the criticisms Latour directs at the Strong Program, the

assumptions he identifies as lying behind it, and the main lines along which hethinks the enterprise should be reformed. His aim is to produce some manner of

non-sociological, non-reductionist analysis of knowledge, one that neither reduces


7/32

87Anti-Latour

nature to society, nor society to nature. He calls this project anthropology but it

is not the anthropology of an EvansPritchard or a Mary Douglas. The entire Durk-

heimian tradition is dismissed on the grounds that it is scientistic and assumes a

modernist division between science and society of the very kind that should be

challenged (Latour, 1990, p. 167).

3. Assessing Latours Criticism

Latours attack depends on a specific characterisation of the aims of those who

pursue the Strong Program. They are said to be trying to explain nature in terms

of society. In reply it has to be said that this is a profound misrepresentation. The

aim isnt to explain nature, but to explain shared beliefs about nature. The enquiry

is into the character and causes of knowledge, or what passes as knowledge, and not

(in general) into the objects which the knowledge is meant to be about. Obviously I

will soon have to confront the question of whether, and how far, we can talk aboutthe knowledge of an object without talking about the object itself, but we need to

begin by getting the overall aims of the enquiry properly in view. This is something

Latour fails to do. The idea that anybody should be trying to explain nature by

society is so peculiar that it is surprising that a critic should ever impute it,

especially given that the formulations of the program are explicit in their material-

ism (see Bloor, 1976). Nevertheless, throughout the entire discussion, Latour makes

no systematic distinction between nature and beliefs about, or accounts of, nature.

He repeatedly casts the argument, his own as well as that of his opponents, in

terms of nature itself rather than beliefs about it. (Remember that for Latour, it is

society and nature, not society and accounts of nature, which are co-produced.) It

is as if he has difficulty telling these two things apart. The quotation above, about

the innovator crossing the boundary between nature and society, is a case in point.

It is, however, easy to see why Latour proceeds in this fashion. He rejects the

subjectobject distinction, and drawing a boundary between nature, and beliefs

about nature, is just a form of this distinction.

Latours criticism, then, starts by ignoring the fact that the Strong Program is

part of a naturalistic and causal enterprise. From the standpoint of the Strong Pro-

gram, society itself is part of nature. The word nature refers to the all-

encompassing, material system in which human animals and the entire pattern of

their interactions, and all the products and consequences of these interactions, have

their allotted place. To talk about society explaining nature, when it is but one part

of nature, is incoherent. Knowledge itself is just one more natural phenomenon.

Rather than positioning the Strong Program at the subjectivist end of the subject

object axis, as Latour does, it would be closer to the spirit of the enterprise to put

it at the opposite end.3

3I am not suggesting that Latour would be any more sympathetic to the Strong Program if he hadfully and properly realised that it was a naturalistic enterprisehe surely would not, given that he isanti-naturalistic and anti-causal. For an example of anti-causal thinking see Latour, 1996, p. 88.


8/32


Because Latour has picked up the wrong end of the stick it isnt surprising that

his subsequent account of the symmetry postulate is confused. That postulate is

not expressive of, or dependent on, an underlying asymmetry of attitude towards

nature and society of the kind he alleges. A correct, naturalistic reading of the

symmetry principle implies that both nature (that is, non-social nature) andsociety will be implicated in the formation of belief. The symmetry to be insisted

upon is that both types of cause, both our experience of the world of things and

the world of people, will be implicated in all bodies of collective belief. Systems

of belief, that is, shared and institutionalised forms of knowledge, are the medium

through which people co-ordinate their shared interactions with non-social nature.

Having some causally structured relationship with the material world is unavoid-

able, and will be compellingly present in all cultures. Adopting a symmetrical

stance means acknowledging both of these dimensions in all cases. Rather than

believing, as Latour thinks, that nobody has access to the real, the position wouldbe better expressed by the slogan: all cultures are equally near to nature. This

means insisting that false systems of belief engage with nature according to the

same general principles as do true ones. However uncomfortable it may be, adopt-

ing the symmetry principle means that Mendelism and Lysenkoism must be seen

as two different ways of causally engaging with (non-social) nature. Both will

involve sensory input from the world, interactions with things and people, the

application of existing cultural resources, and widely shared, though disputed, stan-

dards and goals. Contrary to Latours presentation, the position associated with the

Strong Program points to both Mendelism and Lysenkoism, and both Newtons

mechanics and Einsteins mechanics, and both Fresnels wave theory of light and

Brewsters particle theory, and both Kochs germ theory and Pettenkofers miasmic

theory of disease, being engaged with nature. They all, in their time, had the charac-

ter of social institutions, but that does not mean their practitioners and believers

were not causally interacting with nature. In their different ways, and with different

degrees of success, they were. Somehow the naturalistic and materialistic emphasis

on both has become transformed, in the critics mind, into an idealistic and sub-

jectivist neither.This approach may seem very counter-intuitive. To sharpen the issue let me

concentrate on what is, for the sociologist, the least favourable of the cases listed

above. If we start from the assumption that, say, Mendels theory is true or close

to the truth, while Lysenkos neo-Lamarckian approach is profoundly wrong, and

probably bolstered by false data about crop yields, then how can we sustain a

symmetrical approach? If one corresponds to reality while the other, for various

reasons, fails to correspond with it, how can they both have the same kind of

relationship to it? How can they both be, as the slogan had it, equally close to

reality?We need to examine the connection that holds when something stands as a true

representation of something else. What does the Strong Program have to say about


9/32

89Anti-Latour

the relation of correspondence involved? The point is that, on a naturalistic

approach, correspondence has to be seen as a relation which actors themselves

assert or impute or accept, rather than something operating as a real cause. These

assertions, imputations and acceptances are effects to be explained, rather than

causes which can be cited in explanations. More specifically, corresponding, ornot corresponding to reality, are not causal relationships that bodies of belief bear

to their referent. Beliefs do have causal connections to things in the world, but the

words correspond and do not correspond do not capture those connections.

Neither relationship is a genuine, or naturalistically specifiable connection existing

in its own right. True and false theories do not represent two natural kinds of thing,

any more than a piece of land that I own, and a piece of land I do not own,

constitute two, different natural kinds of land. In both cases we are dealing with

what might be called a moral rather than a causal discrimination.

There is, of course, a causal story to be told as to why discriminations of truthand falsity are made in the way they are, upgrading one theory and downgrading

another. In the MendelLysenko case, like the others, it would be the aim of a

supporter of the Strong Program to tell that story. If it involves mentioning alter-

ation of the data with the intent to deceive, whether by Mendel or Lysenko, then

so be it: that becomes part of the story. In general, the account would deal with

the pragmatics and contingencies of belief which would, for both theories, involve

the generation and processing of data, its selection and evaluation, its perceived

relation to existing bodies of theory, a distribution of expectations and power, and

a set of goals and purposes. All of these judgements and decisions would have tobe anchored in the practices and purposes of the relevant groups. In this case the

groups were, roughly, provincial plant breeders and party bosses on the one hand

and, on the other, metropolitan scientistsoften a generation olderbased in aca-

demies and universities.4

4Given the understandable odium surrounding Lysenko and his supporters, it may seem strange totalk of their engagement with reality in the same breath as that of the geneticists who suffered at theirhands. To offset this strangeness it is worth examining the report on The New Genetics in the Soviet

Union, published under the auspices of the Imperial Bureau of Plant Breeding and Genetics (Hudsonand Richens, 1949). Of course, the writers of this report could not see everything that was going onat the time, but their matter-of-fact survey and analysis of the literature can help us recover the awarenessthat there were genuine intellectual issues at stake behind the brutality. For example, at the time, itwas unclear whether certain empirical results called into question the current understanding of geneticprocesses, or whether the genetic theory could be rescued by blaming the effect on viruses. There wasalso room for argument about the appeal to polygenes, that is, explanations which concerned traitswhich seemed, to the geneticist, dependent on the operation of very large numbers of genes. This wasan important issue because most of the economically significant traits of concern to practical plantbreeders were beyond the powers of explanation of current genetics. To its critics the appeal to poly-genes was just like the introduction of new epicycles to save the old earth-centred astronomy. It is alsoworth remembering that the great success of wheat growers in the United States, standing in such starkcontrast to the problems in the USSR, is itself a phenomenon that needs critical analysis. Their use of

double-cross hybrids is often counted as a triumph of Mendelian genetics, but (a) according to Joravsky(1970, p. 283) they were developed empirically and, at the time of writing (1970) were not understoodtheoretically, and (b) it was not, in any case, possible to transfer this practice to the Soviet Union. Itwas tried and failed because it depended on the (traditionally scorned) need for annual seed purchases,


10/32


Talk about a theory corresponding to reality is simply a vocabulary for express-

ing the upshot of these varied and complicated processes. It conveniently encodes

their outcome, but it doesnt reveal or refer to what they really consist in. For

everyday practical purposes such a convention works very well. It provides the

practical discrimination we want, and draws together a diversity of processes in aconvenient manner. The danger comes when such talk, of true and false, etc.,

is taken out of its workaday context and treated as a given in reflective, analytical

or philosophical enquiries into the working of science. Then it causes trouble by

encouraging simple and misleading pictures, pictures that purport to refer to the

causes of the judgements that we make, when really they are the effects of

those judgements.5

Having now dispelled, I hope, the idea that the Strong Program is part of the

project of explaining nature by society, and having re-asserted its naturalistic cre-

dentials and aspirations, I am in a position to answer Latours main charge. Thisis his idea that social factors are too impoverished to explain the rich detail of

scientific work. In the terms in which he understands these things he is, of course,

right. Unfortunately he is working under the false assumption that, according to

the Strong Program, a scientist is to be thought of as responding to society rather

than to (non-social) nature. We have seen this is not the claim. The working

assumption (unless there is specific evidence to the contrary) is that scientists are

always responding to nature, but doing so collectively through their shared conven-

tions and institutionalised concepts. There has never been any need, or any tend-

ency, within the Strong Program to deny the subtle and detailed character of whatscientists observe, or to deny that it plays a role in prompting and sustaining belief.

Indeed, it serves the purposes of the sociology of knowledge very well to acknowl-

edge the complexity, richness and causal efficacy of sensory input. We can assume

that observation will always enable us to uncover a reality which is more compli-

cated than we can assimilate into our current conceptual schemes and theoretical

systems. Experience and practical involvement with the world will endlessly gener-

ate anomaly. Nature will always have to be filtered, simplified, selectively sampled,

and cleverly interpreted to bring it within our grasp. It is because complexity must

be reduced to relative simplicity that different ways of representing nature are

always possible. How we simplify it, how we chose to make approximations and

selections, is not dictated by (non-social) nature itself. These processes, which are

collective achievements, must ultimately be referred to properties of the knowing

subject. This is where the sociologist comes into the picture.

Explaining the way in which agents negotiate the relation between a complex

as well as the extensive use of weed killers and chemical fertilisers. In modern day parlance what wasneeded was not this (relatively) high technology of the day but appropriate technology.

5

What is needed is a pragmatists sense of the shortcomings of the idea of correspondence, com-bined with a sociological sense of the significance of the category of truth as external, authoritativeand compelling. See Durkheims lectures on Pragmatism and Sociology, partially translated in Wolff(1964) and fully presented in Durkheim (1983).


11/32

91Anti-Latour

(non-social) nature and a heritage of past achievements is not an exercise remotely

resembling the one Latour describes. No one is trying to construct Pasteurs results

out of his conservatism, his Catholicism, his loyalty to the Empress and a few

personality traits. Of course that enterprise is hopeless. It would, in any case, be

a misconceived piece of individualistic thinking rather than a genuine, sociologicalenquiry. The sorts of question that can be asked, and to whose answer the sociol-

ogist can contribute, concern the range of interpretations that might have been put

on Pasteurs observations, the way his questions were framed, and his techniques

for dealing with the uncertainties and unresolved problems in his data. Why did

he bring these particular interpretive resources to bear, and why did he employ

them in this precise way? After all, Robert Koch didnt respond to Pasteurs data

in the precise way Pasteur did, even though they were well informed, expert and

rational scientists who shared many presuppositions about the causation of disease.6

4. The Agency of Things

Latours charge that the Strong Program denies agency to things, because it

reserves all agency and power for social processes, is therefore wrong. The Strong

Program does recognise agency in naturally occurring, non-social things and pro-

cesses, namely causal agency. For example, things have the power to stimulate our

sense organs. Thus Pasteur and Koch could see tiny objects with characteristic

shapes when they used their microscopes to examine the tissues and fluids of dead

animals. Equally obviously, things impinge on us in a mixture of subtle and unsub-tle ways. For example, we would probably fall ill if we injected some of this fluid

into our blood stream. I do not, however, think that such references to the causal

agency of things in nature would impress Latour. He would see in it nothing more

than an oscillation between the two poles, the nature pole and the society pole, of

the framework he has already identified and rejected. The claim made about the

role of objects would, at most, shift his perception of the Strong Program from

being an extreme theory to being an eclectic one which represents knowledge as

made out of social and non-social ingredients.

Why not have such a theory? Latours argument goes like this. The attributionof causal agency to things is itself an exercise of knowledge. In fact it not only

constitutes an employment of the very knowledge the sociologist typically aims to

explain, it also presupposes a division of reality into the categories of the natural

and the social. Now that division is itself the very thing Latour wants to render

6First, Koch thought that Pasteurs work on anthrax was just a repeat of Tiegels earlier experimentsand added nothing new. See Carter, 1988, pp. 5052. Second, for Koch, Pasteur confused anthrax bacilliwith other organisms. This was connected with their divergent opinions about the variability to be

expected within a given species of micro-organism, and this in turn was connected with the differentrelative importance they attached to morphological and physiological characteristics when identifyingmicrobial species. Geison attributes this difference to their different scientific backgrounds and skills.See Geison, 1974, pp. 397399.


12/32


problematic. Any taken-for-granted use of the distinction would thus, in his opi-

nion, be question-begging and lead to a superficial, rather than a fundamental,

analysis. So, for Latour, the appeal to natural causes, of the kind I have just

made, is the outcome of the very process we should be studying (see e.g. Callon

and Latour, 1992, pp. 347348).There is something right about this objection and also something deeply wrong.

What is clearly right is that no causal, naturalistic explanatory program, such as

the Strong Program, can proceed without making some substantial assumptions

about what the world is like. In so far as it does this, its practitioners will already

be making some manner of claim to possess knowledge. The problematic feature

of the objection is whether or not the point at which, and the manner in which,

followers of the Strong Program make their claim to knowledge is any the less

justified than the way others make itwhere others includes the critics, such as

Latour. They also need, and utilise, some manner of knowledge claim in theirwork, so this is not a predicament confined to followers of the Strong Program.

Of one thing we can be sure: nobody can develop any position in a wholly presup-

positionless way. Nobody can turn every resource into a topic without finishing

up with topics which they have no resources for tackling. The difficulty is to decide

which things should be topicalised for investigation and which should be reserved

as resources.

What are the standards for deciding what is legitimate and prudent here? We

shall see in Section 5 that the line taken by Latour is to try to construct a fundamen-

tal ontology and a set of basic philosophical categories for describing events at the

level below that at which social and natural realities have crystallised, that is, at

the origin point of Fig. 2. The description of these events, couched in his philo-

sophical vocabulary, is Latours basic resource: everything else is a topic for inves-

tigation. An alternative strategy, more in keeping with the Strong Program, would

be to adopt an approach loosely derived from the empiricist tradition. The sociol-

ogist needs to have a grasp of what the agents under study are responding to, that

is, what aspects of the world have been disclosed to them in their experience, and

what predicament they take themselves to be in. If we can isolate the stimulusthen perhaps we can begin the task of explaining the response. Of course, the

real concern will not be with individual, psychological responses as such, but with

those responses as mediated by a collective understanding, with its shared traditions

and conventions.

There is no question here of trying to employ the empiricists ideal of a pure

data-language because, for good sociological reasons, such a thing is not available

(see Hesse, 1974). Fortunately we dont need it for the task I am recommending.

It doesnt matter greatly how we specify or describe what the agent experiences,

as long as we manage, somehow, to capture that experience in a way that is suf-ficiently neutral for our purposes. Often scientists do this for themselves when they

are unclear about the identity and causal role of the objects under study. For


13/32

93Anti-Latour

example in 1850 the biologist Davaine reported seeing small filiform bodies about

twice the length of a blood corpuscle in the blood of sheep with anthrax. He

didnt, at that point, say he was looking at the cause of anthrax; it could just as

well have been a symptom or a by-product as a cause (Carter, 1988, p. 43). Davai-

nes description isnt cast in a pure data-language, but it is neutral amongst therange of likely theoretical alternatives (e.g. cause or symptom). Or, to take another

example, we might say that both Priestley and Lavoisier were familiar with a cer-

tain reddish powdery substance prepared from mercury. Concentrating on the

thing as reddish and powdery should encourage us to be puzzled by Lavoisier

calling it mercury oxide, and not just with Priestleys calling it red calx. Pro-

vided the (old) principle of symmetry has a clear hold on our thinking we could

then, if we wished, dispense with the discipline of empiricism and adopt a less

restrained, more realist sounding vocabulary. We could just say there is as much

of a problem why someone should call mercury oxide mercury oxide as to whythey should call it anything else. All acts of naming place the item in a system of

classification, and that system itself must be sustained as a system, as something

shared, a collective achievement going beyond the thoughts within any individ-

uals head.

The important point is to separate the world from the actors description of the

world. It is the description that is the topic of enquiry, and the proposed separation

is one of our resources. This is all just another way of saying we must respect the

distinction between the object of knowledge and the subject of knowledge. For

example, it is often better for the historian of science, or the sociologist, to avoid

saying Robert Millikan observed electrons, or observed the effects of electrons.

That talk should be left to Millikan himself. It is better to say that he observed

something he attributed to, and explained by, a postulated entity he called an

electron. In this way we might be less tempted to think that nature has an automatic

tendency to generate those particular verbal descriptions or responses. If we

believe, as most of us do believe, that Millikan got it basically right, it will follow

that we also believe that electrons, as part of the world Millikan described, did

play a causal role in making him believe in, and talk about, electrons. But thenwe have to remember that (on such a scenario) electrons will also have played

their part in making sure that Millikans contemporary and opponent, Felix Ehren-

haft, didnt believe in electrons. Once we realise this, then there is a sense in which

the electron itself drops out of the story because it is a common factor behind

two different responses, and it is the cause of the difference that interests us. For

this reason, we are bound to pay special attention to the data rather than the

interpretation, that is, to what Millikan and Ehrenhaft actually saw, the readings

they reported and the measurements they entered into their laboratory notebooks.

Cultivating an empiricist sensibility can be a useful tool. Concentrating on whatcan be visually or otherwise sensed sustains our awareness of the gap between

objects and their descriptions. These descriptions provide the main subject matter


14/32


and problem for the sociologist of knowledge. That is what the principle of sym-

metry is meant to keep before our minds.

I now want to make a conjecture. My suspicion is that the critics of the subject

object distinction, such as Latour, think that giving a causal role to nature is tanta-

mount to the assumption that certain descriptions of nature are to be tacitly privi-leged. If this were to be anyones ground for calling into question the subject

object schema, they would be guilty of confusion. They would be inverting the

truth. Only by sustaining the distinction between subject and object, and by driving

a wedge between nature itself and the descriptions of it provided by the knowing

subject, can we highlight the problematic character of those descriptions. It is those

who dont mark the different contributions of the subject and the object who pave

the way to error. They tempt us to think of the transition from the object (under

a given description) to the subjects response to it (in terms of that very same

description) as if it were unproblematicbecause, for them, there is no real tran-sition to be made.

The idea that there is a direct correspondence between the terms of a theory and

entities answering to them in the world might be called direct realism or naive

realism. This is the assumption that if a theorist mentions some type of entity,

and if their theory can be made to work effectively, then its terms must stand in

a one-to-one link to the things mentioned. If the talk is about electrons or microbes,

then there must be electrons or microbes; if the talk is about caloric or lines of

magnetic force, then there must be such a stuff as caloric and such things as lines

of magnetic force. Critics have pointed out that something remarkably like direct

or naive realism turns up in Latours methodology (Collins and Yearley, 1992).

Before looking at that, however, I want to show how it is possible to be a realist

(or materialist) about nature without assuming that any particular theoretical

description of it is uniquely correct. We need to remember that we can be realists

without being direct or naive realists.

Our connection with nature does not depend on each general name in our

theorieseven our successful theoriescorresponding to a natural kind. A system

of knowledge is used, employed and assessed as a whole. If it works as a wholewe are, of course, inclined to project the parts of the theory onto nature. When it

ceases to satisfy us we regain a sense of how varied, complicated, contrived and

contingent those links really are. Obviously individual terms in the theory will

have individual occasions of use. We talk about these electrons, these microbes,

these lines of force, and so on. On those occasions particular experiential episodes

will prompt the application of our terms, but that doesnt mean some uniquely

direct or successful reference has been achieved. The entire system of classification

is implicated and, before too long, this may change. It is best to think of theoretical

systems as a whole having utility and embodying an overall adaptation to reality,where reality is rich enough to permit numerous possible adaptations and numerous

possible descriptions and classifications (see Barnes et al., 1996).


15/32

95Anti-Latour

5. The Latourian Alternative

A number of writers have already responded in a thorough and critical way to

Latours proposals (see for example Amsterdamska, 1990; Collins and Yearley,

1992; Gingras, 1995; Knorr-Cetina, 1985; Schaffer, 1991; Shapin, 1988; Sturdy,

1991; Van den Belt, 1995). Though they begin from different starting points their

evaluations show a remarkable convergence. Latours style is seen as lively and

engaging but his recommendations are treated as unconvincing and his thinking is

judged to be confused. Taken collectively I find that the criticisms in this literature

are devastating. Some of what I have to say in this and the next section unavoidably

overlaps with, and has benefitted from, points made by these critics, but I shall

couch my observations in a way that develops the line of my own, overall argu-

ment.

To begin with, it has to be said that Latour never succeeds in giving a clearaccount of the process he calls the co-production of science and society. Indeed,

such accounts as he does give are deeply obscure. Nor is the situation helped by

his appeal to dark sayings from Serres: Jimagine, a lorigine, un tourbillon rap-

ide. (Latour, 1990, p. 163) (On this basis Latour sometimes embellishes the origin

point of Fig. 2 with a small spiral motif.) I shall illustrate these deficiencies by

Latours own words. First recall Latours diagram, and the vertical axis which is

meant to signal the stability with which social groups mark the distinction between

nature and society. Let us ask whether, according to Latour, Pasteurs microbes

were really there in nature all along, waiting to be discovered, or whether they arejust an idea invented by Pasteur, which caught on (a pure social construct)? Latour

says we should not ask this question. We must neither be realists about microbes,

nor try to reduce Pasteurs science to social conditions. Properly understood this,

or some version of this, may be the right answer, but let us see what Latour takes

to be its appropriate, practical expression. He says:

I want to stress again that I am not interested here in offering a social or political

explanation of Pasteur or an alternative to other cognitive or technical interpretations.

I am interested only in retracing our steps back to the moment when the very distinc-

tion between content and context has not yet been made. (Latour, 1988, p. 252)

The word context here obviously refers to social context, so content, that is,

the content of knowledge, is what Latour elsewhere calls the object of knowledge.

What, then, confronts the analyst at the moment before the distinction between

content and context, or subject and object, has been made? At this moment, it

seems, we are to think of ourselves as dealing neither with microbes (which would

fall under the category of content, or the object of knowledge) nor with a society

responding to the microbes it knows about (this would fall under the category of

context, or the society as knowing subject). We are, rather, dealing with microbesin the making and with a society in the act of making them, and in the act of

making itself. At this point:


16/32


it is crucial to treat nature and society symmetrically and to suspend our belief in a

distinction between natural and social actors. (Latour, 1988, p. 260)

Whereas followers of the Strong Program would recommend treating nature and

society symmetrically by saying that both have causal efficacy in bringing about

belief, notice that Latours generalised symmetry refers not to two causes, butto two effects. Nature and society are two effects with a common cause or, since

Latour is critical of causation, two processes with a common basis. The crucial

phrase, and the one that is characteristic of Latours recommended approach, is

that of suspending belief in the distinction between natural and social actors.

What does this mean? Well, Pasteur was a social actor and his microbes were

natural actors, so we have somehow to put aside all our usual assumptions about

how different they are. Similarly, Millikan and Ehrenhaft were social actors and

the (presumably real) electrons of the one, and the (presumably non-existent) sub-

electrons of the other, would be natural actors. Again we must suspend our belief

in this distinction.

Astonishing though Latours suggestion may seem we ought not to be too ready

to scoff at such a goal. After all, it is deemed philosophically respectable to argue

that minds are brains and that brains are computers. Such positions may be rejected

by their critics as mistaken, but they are taken seriously. Formulations of the Strong

Program have also been couched in terms of a background philosophical material-

ism, and, for the materialist, humans such as Pasteur and Millikan are just like

microbes, or any other material object, in being collections of electrons and other

basic particles. At some, ultimate, metaphysical level many of us are going to find

ourselves in a posture that is not too dissimilar to Latours. The important point

is how these highly general themes find an expression in methodology, for instance,

in the handling of empirical material drawn from the history of science. It is here

that the real oddity of Latours position becomes significant.

Latour wants to bring these ultimate issues right into the foreground. They are

not distant goals, referring to some future synthesis. Unlike the identification of

brains and computers they are to be given direct expression in the analysis of the

day to day conduct of science. This is why he wants the analyst to operate, notwith a sociological vocabulary, but at a level at which the ordinary categories of

person and thing are held in suspense. Another, more abstract, vocabulary must be

brought into play. Such a vocabulary, Latour is prepared to concede, doesnt yet

fully exist, but it is under construction and further work of this kind should be one

of the priorities of the field (Callon and Latour, 1992, p. 354). In the meantime

we can appeal to the philosophical tradition for a range of neutral and monistic

concepts. Thus, instead of people, like Pasteur and Millikan, and things like

microbes and electrons, we are to deal with what Latour calls entelechies or

monads or quasi-objects or forces. The processes which drive everything alongare to be conceived abstractly as alignments of forces, or oppositions of forces, or

trials of strength. Nevertheless, Latour adds the significant qualification:


17/32

97Anti-Latour

No, we do not know what forces there are, nor their balance. We do not want to

reduce anything to anything else. (Latour, 1988, p. 156)

and

In place of force we may talk of weaknesses, entelechies, monads, or more

simply actants (Latour, 1988, p. 159).

The reference to Leibnizian monads seems to be meant seriously. Recall that,

for Leibniz, monads are said to be windowless. Latour tells us in so many words

that in his own system there is no question of an external referent (Latour, 1988,

p. 166). Reference is always internal. Thus:

Every entelechy makes a whole world for itself. It locates itself and all the others; it

decides which forces it is composed of; it generates its own time; it designates those

who will be its principle of reality. (Latour, 1988, p. 166)

What are we to say to this? The main point to be made, and made emphatically,is that it still remains wholly unclear how to connect this metaphysical talk to

historical and everyday reality. It exists merely as a fantastic gloss on a body of

fact, such as Pasteurs work in microbiology, that exists quite independently of it.

We are told to encourage the new perspective by deliberately inverting our usual

conceptual conventions, using a purposive vocabulary for things which dont have

purposes, and a mechanistic vocabulary for things that do. We must try to think

of Pasteur as if he were a microbe, and microbes as if they were like Pasteur, or

treat Millikan as if he were an electron and electrons as if they were like Millikan.

But unless we are very confident indeed that the exercise is necessary and justified,this looks like a formula for imposing confusion on ourselves: it is obscurantism

raised to the level of a general methodological principle.7

Here we need to tread carefully, because Latour runs together general metaphys-

ical claims with specific issues thrown up by particular historical cases. He points

out that Pasteurians frequently spoke of microbes as if they were like people, talk-

ing about them as one would an enemy, or as an unwelcome guest or as a

murderer with a deadly mission. As actors categories such talk must be taken

as we find it. If it turns out to be more pervasive and literal in intent than we had

expected, then so be it. Much of Latours position is derived from the specific case

of the Pasteurians, and in so far as it carries any general methodological message,

corresponds to the widely accepted strategy that an investigation must start from

a grasp of the actors point of view. Here, however, actor means human actor,

and yet it is this very distinction Latour wants to break down. On the one hand

Latour will speak of microbes as literally having interests (e.g. It uses your inter-

ests to carry out its own, Latour, 1988, p. 33). On the other hand, we are told

that he is using the idea of agent in a broad, indeed all-inclusive, semiotic sense

7More recently Latour has appealed to Whiteheads metaphysics as a vehicle to convey his point(Latour, 1996). I do not believe this has advanced the argument in any way.


18/32


(Latour, 1988, p. 35), which seems to be a way of telling us not to take such

talk literally.

Unresolved tensions of this kind are endemic in Latours text as he struggles,

unsuccessfully, to convey what he means by the co-production of collective things

(Latour, 1992, p. 287). Many of the specifications he gives to the enterprise arenegative: he is not, he insists, merely saying that things are half natural, half

social. They are neither objects, nor subjects, nor a mixture of the two (Latour,

1992, p. 282). But when it comes to a positive specification we find that the langu-

age, in so far as it conveys anything, begins to slip back into the more familiar

language of the sociology of knowledge. After baffling talk about quasi-objects,

which are produced and circulate, we hear that they are a new social link that

redefines at once what nature is made of and what society is made of (Latour,

1992, p. 283). Here is something we can grasp: it is a social link we are dealing

with. Elsewhere it appears that co-production is a process that resides in commonpractice (Latour, 1992, p. 281), and that objects and subjects are belated conse-

quences of an experimental and historical activity (Latour, 1992, p. 284). This,

again, is something we can hang on to. No doubt these terms are meant to have

all manner of subtle overtones which I am here passing over, but the aim is to

salvage something concrete and usable from the obscurity. Monads and entelechies

having been left behind, we are left with (something like) social links, social prac-

tices, historically situated activity, and even familiar sounding cultural categories

such as the experimental. But if we have social links, in anything like the usualmeaning of the words, we must have a society. If we have practices, in anything

like the usual sense, we must presuppose a form of social life. Latours talk about

making nature and making society, it seems, cant be taken too seriously. Really

it presupposes a nature and a society all along.

Latours attempt to get to metaphysical bedrock doesnt work: he cant get away

from a pre-existing nature and a pre-existing society. He is brought back to the

same starting point as the sociologist of knowledgethat boring creature who Lat-

our berates for not getting beyond Kants Critique. It seems that, after all, we have

to begin our investigations into the nature of knowledge from where we are stand-ing. Our feet are on the ground of nature, and our position is in the midst of an

existing culture, our own culture. This is, perhaps, not quite as limiting as it may

seem, or as Latour paints it, because we dont have to take that culture entirely at

face value, or respond to it uncritically. Nevertheless, such critical distance as we

do achieve can only be got by using the resources of that culture itself, using one

bit of it as the basis for looking at another bit. Given that our culture is complicated

and pluralistic, and equipped with a sense of its own history, and divided into

opposing traditions, these resources for achieving the necessary role-distance are,

I suggest, as rich as we are ever likely to need.


19/32

99Anti-Latour

6. The New Symmetry in Practice

So much for Latours attempt to spell out his methodology on the theoretical

plane. What does it look like in practice? The answer is that it looks suspiciously

like ordinary sociology of scientific knowledge, albeit of a rather limited and one-

sided kind.8 The only difference is the addition of some obscure terminologicaltwists, and repeated assertions to the effect that the two enterprises are disjoint and

opposed. Consider, for example, his main case-study dealing with the reception of

Pasteurs work. We are explicitly advised against doing what any sociologist of

knowledge would do, namely, identify different groups, locate their interests, and

see if their differential response to a claim might be rendered intelligible in these

terms. This old-style reductionist sociology is, we are told, obsolete (Latour,

1988, p. 256). Latour has asked sociology to abandon its social groups and its

interests and allow the actors to define themselves (Latour, 1988, p. 51). But as

reviewers of The Pasteurization of France were quick to point out, if we look at

what Latour actually does, we find him conforming exactly to the procedure he

has just denounced. He identifies social groups and their interests and depends

entirely on these to tell the story of the response to Pasteurs work. He identifies

the hygienists, the army doctors, the surgeons, and the physicians. He shows how

Pasteurian techniques suited the purposes of the first three of these groups but

not, initially, those of the physicians, who feared disruption of the doctorpatient

relationship. The emphasis on preventive rather than curative techniques posed a

potential threat. Hygienists on the other hand, who had been frustrated by their

lack of success in predicting and preventing outbreaks of disease, found in Pasteurs

techniques an effective vehicle for the pursuit of their aims. This is how Latour

makes sense of their enthusiastic endorsement of Pasteurian ideas. We are told:

The immense trust in Pasteur derived partly from the work that he had done before

1871, which did not concern infectious diseases, and partly from the social movement

that needed these discoveries but went well beyond them without waiting for them

to be made. (Latour, 1988, p. 30)

The social movement in question is that of the hygienists. They went beyond

Pasteur by immediately responding to his work on a limited range of diseases asif it held the key to all diseases. But the central point is the reference to an identifi-

able social group which needed these discoveries. Further, Latour says of the Pas-

teurians:

Working in few laboratories, they pronounced words that were immediately regarded

as truthful and were integrated into evidence that at last allowed the hygienist move-

ment to get on with its work. (Latour, 1988, p. 34)

8As Schaffer points out, Latour is profoundly asymmetrical as between the Pasteurians and theiropponents (Schaffer, 1991, p. 185).


20/32


So not only do we have a need, we have a group with a conception of its work

that it can see how to further. This provides us with all the ingredients for the

identification of interests, and their explanatory imputation to a social group.

Perhaps Latour senses the danger of this reading, because he tries to forestall

it. As well as the passage already quoted, asking sociologists to abandon theirappeal to social groups and interests, he says:

Once again, whenever I use the words interest and interested, I am not referring

to the interest theory expounded by what is now called the Edinburgh School I

am rather referring to the notion of translation Interest means simply what is

placed in between some actor and its achievements. (Latour, 1988, p. 260)

Despite the denial, if we check this claim by trying to substitute the words in

between, or the general idea of in-betweenness, in the passages in which Latour

uses the word interest we find it does not work. For example in The Pasteurization

of France we read:

Either the physicians could use what was taking place in the Institut Pasteur to

advance their own interests, or they could not. (Latour, 1988, p. 120)

and:

The Pasteurians added to society a new agent, which compromised the freedom of

all other agents by displacing all their interests. (Latour, 1988, p. 122)

and again:

Now the doctors, after many other groups, by giving Pasteurism a push would also

advance their own interests. (Latour, 1988, p. 127)

All of these passages make perfect sense if the word interest is read in the

standard way to refer to a benefit the group would gain if some course of action

were pursued. The idea that it refers to something quite other than this, something

having the role of coming between an action and its achievement, is not readily

intelligible. Instead of admitting openly that he is, after all, in the business of

giving run-of-the-mill interest explanations, Latour simply makes the same points

but transposes them into another vocabulary. For example:

If hygienists had wanted to open up a dispute, they could have done so. The absence

or presence of a controversy is a measure only of the angles of movement of the

actors. (Latour, 1988, pp. 5253)

This could just as well be expressed by saying that the hygienists might have

found Pasteurs claims open to argument had they been inclined to do so, and they

might have been so inclined if it had served their interests. Instead of being told

about the perceived coincidence of the interests of the hygienists and the inner

group of Pasteurians we hear about their angles of movement. I do not want toquibble over terminology, but do these metaphors really enable us to say anything

deeper, different, or better than standard talk about interests? I think not.


21/32

101Anti-Latour

7. Relativism

In developing his criticisms of the Strong Program, Latour seeks to distance

himself from a position he calls relativism. Relativism, as he presents it, is said

to be a direct consequence of the (first) symmetry principle. Given that there are

a variety of different positions that might be called relativism it is important toidentify exactly what Latour is denouncing. I shall follow the discussion in Science

in Action, though this is entirely representative of his treatment elsewhere. First,

we must notice that Latour follows the widespread trend of treating relativism

as a contrast to realism (rather than, as should be the case, as a contrast to

absolutism). Thus realists are said to believe that scientific controversies about how

best to represent nature are settled by nature itself, while relativists are said to

believe that Nature will be the consequence of the settlement (Latour, 1987, p.

99). (Notice that Latour says nature not beliefs about nature.) Second, relativ-

ism is taken to be an evaluative position. Relativists are said to be committed to

defending bodies of belief against various charges, such as the charge of irration-

ality. Their aim, allegedly, will be to convince us that such a negative evaluation

is unfounded or impossible to sustain, and that the body of belief in question can

be defended on the grounds that it is really rational after all. Latour draws a legal

parallel. Relativists, he says, are like defence lawyers, arguing for the innocence

of their client. Whenever the scientific community rejects a theory, as they rejected

phlogiston or caloric or Newtonian mechanics with its absolute space and time,

the relativist must make the case for the defencein the teeth of the scientific con-

sensus.

Every time a charge of irrationality is filed, relativists argue that it is only an appear-

ance that depends on the jurys relative point of viewhence their nameand they

offer a new perspective from which the reasoning appears straightforward. Their pos-

ition is called symmetric (Latour, 1987, p. 195)

Latour finds this position indefensible because, by their commitment to the sym-

metrical idea that all opinions are equally worthy of credit, relativists ignore the

obvious fact that scientists themselves work hard to establish asymmetry, that is,

to make some theories more credible than others. Thus he complains:for four chapters we have followed scientists at work who strive to make their claims

more credible than those of others. So if this enormous work makes no difference

they have wasted their time, I have wasted my time, the readers have wasted their

time. (Latour, 1987, p. 196)

The work that goes into achieving differential credibility takes the form of estab-

lishing alliances, connections, and the enrolment of others, that is, in the creation

of what Latour calls networks. These include all facets of the situation, material

and social. Unfortunately, says Latour, these networks are neglected by relativists:

But in the symmetric stand it is the very existence of the scientific network, of its

resources, of its ability to sometimes tip the balance of forces, that is utterly ignored.

(Latour, 1987, p. 196)


22/32


It would be difficult to imagine a more serious charge against the relativist soci-

ology of knowledge of the Strong Program than that it utterly ignored the

phenomenon of interaction that goes into forming social networks, or that it ignores

the possible role of sensory input in tipping the balance. The charge is, however,

wholly misconceived. The relativism Latour rejects is quite distinct from therelativism of the program he takes himself to be attacking.

First, the relativism of the Strong Program is not to be counterpoised to realism.

As I have emphasised, (non-social) nature plays a central role in the formation of

belief, though how nature is experienced cannot provide a sufficient causal expla-

nation of how it is subsequently described. Second, Strong Program relativists are

not like lawyers trying to make out a case for innocence. If we go along with the

legal comparison they would be better likened to philosophers of law who argue

that there are no absolute standards of justice, or no absolute rights against which

legislation may be assessed. This is a quite different image.The point of the symmetry postulate is to enjoin sociologists to draw back from

making first-order judgements. The point is to make such judgements the objects

of enquiry. It is precisely judgements of this kind which are to be explained. Such

a position is relativist because there are no absolute proofs to be had that one

scientific theory is superior to another: there are only locally credible reasons. Of

course the phenomenon of differential credibility is real. The aim of a relativist

sociology of knowledge is not to ignore or deny such variation, but to explain it.

Latours idea that Strong Program symmetry means saying that all beliefs are equ-

ally credible is wrong. The claim is that all theories and beliefs equally face the

problem of credibility, and hence that all differences in, and degrees of, credibility

are equally in need of causal explanation.

Even if it is accepted that symmetry doesnt imply equal credibility, isnt there

still something right about Latours characterisation of relativism? Suppose a soci-

ologist of knowledge were to examine, say, Robert Kochs early claim to have

identified the bacillus which causes anthrax. At the time Kochs critics said he had

not strictly proven that it was the bacillus alone, rather than some constant concomi-

tant of the bacillus, which is the cause. Koch responded by saying it is impossibleto provide complete and total proof, and therefore meaningless to ask for it. Later,

in his work on tuberculosis, he endorsed, and claimed to have satisfied, what later

came to be called Kochs postulates for establishing causality. These seem to

demand exactly what he earlier denied: the complete isolation of a bacillus so that

it, and it alone, can be known to be the cause of a disease. Has Koch contradicted

himself? Was his earlier claim logically at odds with his mature methodology?

Now, a sociologist of knowledge may well be reluctant to make such judgements.

This looks as if it fits Latours stereotype that, for the relativist, everybody is

innocent of any scientific crime of which they may be charged, in this case, thecharge of logical inconsistency.

In fact something quite different is going on. The aim is not to substitute alterna-


23/32

103Anti-Latour

tive evaluations (innocent rather than guilty, consistent rather than inconsistent)

but to avoid evaluations being used as substitutes for more searching enquiries.

The point a sociologist of knowledge would insist on is that Kochs postulates for

identifying causes must always be applied within a context, that is, against some

background which is being taken for granted. Like all rules and principles, theydo not have any intrinsic basis for their application but always, and necessarily,

depend on local contingencies. Without such a background they would yield no

determinate answers. The later use of Kochs postulates, and the implicit admission

that they did not make impossible demands, took place once the germ theory had

become the effective paradigm for research in the area. The earlier claim, that they

were impossible to satisfy with complete rigour, was made in the context of argu-

ments with critics, some of whom did not yet fully accept the germ theory. In this

context nothing need have counted as satisfying them. The critic could always say

that it was possible that a concomitant cause had been overlooked. In such a contextKochs postulates always yield the result that no cause has been isolated. It is only

in the context of the germ theory as an institution that the postulates discriminate

between what counts as a rigorous proof and what, by comparison, must be dismis-

sed as sloppy reasoning.9

It may well be reasonable to suspect that Koch was inconsistent, or that he

changed his mind without admitting it. The important point for the relativist is that

such an evaluation can only be made once we have established the background

against which the content of his thoughts can be identified. The aim therefore is

not, as Latour thinks, to make Koch out as innocent. There is no such commitmentor preconception. The point is to insist that the logical content of an argument, a

claim, or a beliefthat is, the preconditions of both consistency and inconsist-

encycannot even be properly brought into view, let alone assessed, until they

are relativised to their social context.10

8. Similarities and Differences

Much of Latours new perspective simply shadows things that have already

been argued by sociologists of knowledge. For example, Latour insists that hewants to let actors define themselves. On at least some of its possible interpret-

9Significantly, once the germ theory of disease had become taken for granted the practical level ofproof that was felt to be acceptable fell distinctly short of the satisfaction of Kochs postulates (seeEvans, 1987). The extent to which they need to be satisfied, and what actually counts as their satisfac-tion, is still under dispute as can be seen from contemporary disputes over AIDS (e.g. Cohen, 1994).I should like to thank Henk van den Belt for drawing my attention to these references.

10I have based this example on two papers by K. Codell Carter, The KochPasteur Debate onEstablishing the Cause of Anthrax (Carter, 1988) and Kochs Postulates in Relation to the Work of

Jacob Henle and Edwin Klebs (Carter, 1985). Since I am using the material to make a general pointabout methodology I have not, of course, tried to reproduce the historical detail. A reader of thesefascinating papers will find that the more detail is brought in, the stronger the methodological pointbecomes.


24/32


ations sociologists of knowledge have long subscribed to this principle. The Strong

Program can be said to have adopted this approach by discouraging the analyst

from dividing agents into two, evaluative categories, namely those who subscribe

to what we take to be true beliefs on some subject matter, and those who dont.

Such a division would indeed be a case of not letting the actors define themselves,because we would be imposing our evaluation of their situation on them. But that,

of course, is exactly what the symmetry principle forbids.

Again, Latour doesnt want to assume that interests are stable or that groups can

be endowed with explicit goals (Latour, 1988, p. 260). Fine, but no sociologists of

knowledge are obliged to assume such a thing. Whether interests are stable or

unstable, and whether or not goals are made explicit, are matters for empirical

investigationjust as they are for Latour himself. Though he rightly doesnt

assume that interests are stable, he certainly finds that some of them are, as we

can see from what he says about the hygienists (see for example the commentabout several generations, Latour, 1988, p. 62). Similarly, Latour doesnt want

to say that Pasteurs interests fitted those of the hygienists, but that there was

room for a negotiation about the meaning of contagion (Latour, 1988, p. 255). Is

there any problem here for the sociologist? Are we to believe that sociologists are

not at home with the idea of negotiation? Is the only available account of interests

one in which the idea of negotiation can find no place? Clearly not (see Shapin,

1982). This is simply a hostile characterisation of the sociological enterprise

designed to heighten the impression that it is distinct from Latours approach.

Latours recommendations dont always simply shadow those of sociologists.

One respect in which they differ concerns the relationship between the analyst and

established bodies of scientific knowledge. Here is the problem. The sociologist is

committed to identifying the conventional aspects of knowledge. If something is

conventional then, in principle, there must be a viable alternative. Driving on a

certain side of the road is conventional because we could, in principle, drive on

the other side. Given that we drive on one side, a change in convention might be

costly and impractical, but that doesnt destroy the conventional character of the

practice. Demonstrating a conventional component in knowledge means showingthat understanding could have taken another route without overriding the normal,

biological functioning of the human brain. At any given time such an alternative

might be costly, but that is not the point. The point is that it must be rationally

possible. This can present a difficulty when dealing with up-to-the-minute, and

highly esoteric, knowledge of a kind which currently commands a consensus

amongst the experts. Alternatives are not going to be easy to come by. To produce

them would involve beating the experts at their own game. By its very nature this

may be a practical impossibility. Any alternative that might be suggested is likely

to be dismissed by the experts, on convincing grounds, as inadequate or erroneous.There will then be no known alternative, except error, so it will be impossible to

exhibit the conventional character of the knowledge. Under these circumstances it


25/32

105Anti-Latour

will be tempting to feel that the sociological approach has met its limits, not just

its practical limits, but its theoretical and logical limits. The expert knowledge

under study, we might feel, cannot be dependent in any significant way on conven-

tion. It must correspond directly to reality because the facts of the case give us

no alternative.Under these circumstances a supporter of the Strong Program should stand firm

and go back to the basic principles of relativism. It could well be that a body of

knowledge depends significantly on convention without our currently being in a

position to demonstrate it directly in the particular case. The limitation could be

entirely contingent, for example: lack of ingenuity and creative imagination in the

relevant field. It is then necessary to depend on inductive arguments drawn from

historical cases. Given historical distance, it is easier to show that there are alterna-

tive ways of understanding the data. Copernicus shows us how Ptolemys data

could have been understood otherwise, Lavoisier shows us the alternative toPriestley, Einstein the alternative to Newton, Cauchy and Weierstrass show us the

alternative to the infinitesimals of Leibniz, and Robinson shows why there might

be infinitesimals after all, and so on. The inductive generalisation from such cases

to the consensus of today shows that it too will have alternativesunless someone

can produce remarkable and cogent reasons for thinking that qualitative changes

have suddenly taken place in the nature of knowledge. It is no use a critic of the

sociology of knowledge pointing to the success of current knowledge. That doesnt

represent a qualitative change or a proof that conventionality has suddenly been

transcended: accepted knowledge always works, it always has its successes, untilwe find something that seems to work better.

Latours response to the problems of analysing authoritative and current bodies

of knowledge is quite different from that sketched above. The sociological demon-

stration of conventionality requires that, in a certain sense, the analyst may need

to know more than the social actors themselvesin the way in which historians

need to know more than the historical actors they describe. And, of course, this

may be unattainable in practice. Latour, by contrast, says the analyst does not

need to know more (Latour, 1988, p. 10). His view is that an analyst should only

adopt a relativist stance during periods of scientific conflict or indecision. During

times of scientific consensus analysts should comport themselves as realists. This

is because Latours aim is merely to travel along with the scientists, to follow them

around and describe their opinions and attitudes. Thus:

We do not try to undermine the solidity of the accepted parts of science. We are

realists as much as the people we travel with But as soon as a controversy starts

we become as relativist as our informants. (Latour, 1987, p. 100)

When in the realist mode the analyst is enjoined to take nature as the cause of

accurate descriptions of herself.We cannot be more relativist than scientists about these parts and keep on denying

evidence where no one else does. Why? Because the cost of dispute is too high for


26/32


an average citizen, even if he or she is a historian and sociologist of science. If there

is no controversy among scientists as to the status of facts, then it is useless to go

on talking about interpretation, representation, a biased or distorted world-view

Nature talks straight, facts are facts. Full stop. There is nothing to add and nothing

to subtract. (Latour, 1987, p. 100)

The issue is not however, as Latour presents it, one of trying to undermine

the solidity of science. Demonstrating, say, the underdetermination of theory by

experience, and the negotiability of scientific concepts and conclusions is not, in

any real way, to undermine science. It might undermine a range of theories about

what its solidity consists in, but that is quite another matter. Nor is the issue one

of denying the evidence. While nature may appear to talk straight to the believer,

that appearance is false, and it is just as false during periods of stability as during

periods of instability. Direct realism may describe how things seem to the

believerthe believers phenomenologybut if we are to understand that phenom-

enology, and its variations, we cannot just endorse the agents own perception of

things.11 Latour notwithstanding, there is always something to add to what scientists

say in the conduct of their professional roles. That something is a general model

of knowledge. It is simply untrue to say that sociologists cant be more relativist

than the scientists under study. They certainly can be, and not by denying evidence,

but by taking into account more evidence, namely the evidence from the history

of science, which points to the possibility of alternative understandings. What is

true is that sociologists may not be able to exhibit these alternativ

bloor anti latour

Documents